This Mentored Research Scientist Development Award will support the candidate in establishing an independent research career as a methodologist who develops cutting-edge statistical methodology with direct applications in alcoholism and substance abuse prevention. Career development training areas include: 1) Development psychopathology; 2) Alcoholism and substance abuse prevention research and methodology; 3) Modern computational techniques; and 4) Modeling longitudinal data using nonparametric regression. The specific aims of the proposed research plan are: 1) To develop a new Bayesian latent variable model for assessing dynamic childhood risk for alcoholism in multiple dimensions; 2) To develop a new semi-parametric regression model for validating the association between childhood risk and adolescent substance use development; and 3) To apply the methodologies developed in Specific Aims 1 and 2 to test hypotheses based on the data collected from an ongoing 21-year long prospective study of families at high risk for alcoholism. The methodology developed in this study will be useful in establishing construct validity of commonly used measures of childhood risk. Two statistical programs will be developed and distributed to prevention scientists. The derived longitudinal model based on the empirical data should be useful in projecting a new child's developmental trajectory of adolescent substance use based on his/her childhood risk measures. Thus, it has the potential to provide crucial information about the timing and dose an individual child needs for early intervention.